Weather data and animal module

[bianatur]

Hello, I am running simulations under different weather scenarios over multiple years.
I started comparing arid and temperate weather example datasets using the same herd configuration from 2013 to 2019. We expected to see differences in milk production outputs. However, the results were exactly the same even when changing farm location (county in the config file).
I wanted to check with RuFaS developers whether the changes in weather parameters directly affect the animal module and, therefore, milk production outputs. For example, can RuFaS simulate the impacts of heat stress on milk production?

Thanks,

[allisterakun]

Hi @bianatur,

Your observation is expected behavior with the current model design. The weather data does not propagate to milk production, so identical milk outputs across arid vs. temperate runs (same herd config, same seed) is what the code should produce.

Here is why, with the relevant code:

1. Milk yield is exogenous. Each cow's daily milk comes from a Wood's lactation curve, l * dim^m * exp(-n*dim) where l, m, n are the Wood's parameters, a function of days-in-milk and the animal's lactation-curve parameters plus a small random daily variance. It does not respond to feed quantity, feed quality, ration nutrition, or ambient temperature. See calculate_daily_milk_production in RUFAS/biophysical/animal/milk/milk_production.py, lines 238-265.

2. Weather feeds the crop/soil and manure modules, which affect on-farm feed production and inventory. When home-grown feed is short, though, the ration falls back to purchasing feed off-farm, so a weather-driven change in feed supply mostly shifts the home-grown vs. purchased balance rather than changing what the animals are actually fed.

3. The only mechanism that changes milk is a discrete milk-production reduction that triggers when a ration cannot meet a pen's nutrient requirements, and that only happens if the shortfall cannot be covered by purchasing (i.e. purchase limits are reached). See the reduction logic in RUFAS/biophysical/animal/pen.py, lines 1243-1259 and lines 1332-1344. Under typical inputs this does not occur, so milk is unaffected.

4. On temperature specifically: ambient temperature is passed into the nutrient requirement calculation (see RUFAS/biophysical/animal/herd_manager.py, line 1634), but it does not reach milking cows. Under the NASEM standard, which is the default in all example configs, temperature is not even passed to the calculator (RUFAS/biophysical/animal/animal.py, line 2666). Under NRC, the temperature term 0.0007 * (20 - previous_temperature) is applied only to heifers' maintenance energy, not lactating or dry cows (RUFAS/biophysical/animal/nutrients/nrc_requirements_calculator.py, lines 226-232), and even then it adjusts a requirement, never the milk curve.

So to your direct question: no, RuFaS does not currently simulate heat-stress or cold-stress impacts on milk production. There is no THI or heat-stress relationship on the lactation curve at this time. It is a reasonable enhancement to consider, but it is not in the model today.

Hope that clarifies the identical results. Happy to dig further if you would like to see where feed shortfalls do show up in the outputs, such as purchased-feed quantities and costs.

[bianatur]

Thank you for the very clear reply. It would be really helpful if you could help me see where feed shortfalls appear in the outputs.